Antoni Sort

Built and launched an AI-native career platform from initial discovery and JTBD validation to public beta, onboarding 50+ users and owning the end-to-end product, design, and AI architecture.

AI Unit Economics & Evals: Slashed LLM cost per turn by 330x ($0.20 → $0.0006) with zero quality regression across a 30-question evaluation benchmark by orchestrating narrow task-specific agents and per-profile semantic caching, making a free public AI tier commercially viable.

Designed a DAG skills catalog (Directed Acyclic Graph) fed by live job postings that keeps market supply and demand mapped and embbeded, the backbone for features like evaluation and career roadmaps.

Engineered a harness framework tailored to the product: new AI features ship in 3-6 hours and get tuned daily from telemetry. Traded flexibility for speed, rigor and cost.

Co-led the strategic company pivot alongside the CEO: built the initial functional POC, led the commercial product pitch that secured the FC Barcelona official partnership, and negotiated the deliverable MVP scope.

Led a 12-person cross-functional team: (Eng, UX/UI, Art, Data) to ship the MVP in a fixed 6-month window: 150k downloads 8 months post-launch, and the designed core loop for the app resulted in 2x the industry average for mobile D30 retention in the same app category.

Scalable Platform Monetization: Architected the core digital card economy and progression system to align incentives across three stakeholders (company, sports club, and fans), creating a repeatable white-label product model that closed 8 additional club partnerships.